2 Leonel Valbom (Editor)!!!!!! ARTECH 2010 Envisioning Digital Spaces Proceedings of the 5th International Conference on Digital Arts 22 & 23 April, 2010, School of Architecture University of Minho, Guimarães, Portugal
4 ! ARTECH th International Conference on Digital Arts, 22 & 23 April, 2010 UM, Guimarães, Portugal Preface We are pleased to present the Proceedings of ARTECH 2010, 5th International Conference of Digital Arts, held April, Guimarães, Portugal. ARTECH 2010, as the previous editions, aims at bringing the scientific, technological and artistic community together, while promoting the interest in the digital culture and its intersection with art and technology as an important research field, a common space for discussion and exchange of experiences. ARTECH is now joining researchers, artists and multidisciplinary teams from all over the world and during two days they have the opportunity to share ideas and their work. This year, we had around eighty works submitted, including full papers, short papers and art installations and fifty of them were accepted. All of the papers have been peer reviewed by at least two experts. The proceedings contain the papers of the Artech 2010, 5th International Conference and are divided in 3 sessions: full papers, short papers and art installations, each of these sessions divided by language. The collection of papers represents a wide range of topics such as: Art and Science Theory, Audio-Visual and Multimedia Design, Digital Fabrication, Electronic Music, Generative and Algorithmic Art, Immersive Art, Interactive Systems for Artistic Applications, Media Art history, Mobile Multimedia, Net art and Digital Culture, New Experiences with Digital Media, Technology in Art Education. Many of the papers cover integration of technology in art, other describes artworks with built in technology and some others try to explain processes and techniques. We are grateful to the support of all of the authors and the Scientific Committee who played significant role in the review process, to Keynote Speakers and to Steering Committee of ARTECH. We are also grateful to the sponsors of the conference for their kind support and to the supporting institutions for distributing information. Guimarães, April, 2010 Leonel Valbom President of Program Committee Pedro Branco Chairman of Local Organization! i
6 ! ARTECH th International Conference on Digital Arts, 22 & 23 April, 2010 UM, Guimarães, Portugal! Organization Program Committee Leonel Valbom (President) Ana Soler Baena (Vice-President) Marcelo Wanderley (Vice-President) Scientific and Art Committee Adérito Marcos, UAberta (PT) Álvaro Barbosa, EA-UCP (PT) Ana Amélia Carvalho, IEP-UM (PT) Ana Luisa Roddrigues EA-UM (PT) António Sousa Dias, CICM Paris 8 (FR) Chris Chafe, CCRMA-Stanford (USA) Daniel Tércio, FMH-UTL (PT) Elizabeth Carvalho, CCG (PT) Ernesto Melo e Castro, ESAP (PT) Heitor Alvelos, FBA-UP (PT) Henrique Silva, Bienal de Cerveira, (PT) Ian Oakley, UMA (PT) Ido Iurgel, CCG (PT) Isabel Soveral, UA (PT) James Faure-Walker, University of Arts, UK João Alberto Correia-ESGallaecia (PT) Jon McCormack, Monash U. (AU) José Bidarra (UAb) José Pinto Duarte, FAUTL (PT) Karla Schuch Brunet - UFBA (BR) Laura Baigorri, FBA-UB (ES) Lola Dopico Aneiro, FBA-UV (ES) Lucia Santaella Braga, PUC-SP (BR) Manuel Gamito, Sheffield U. (UK) Marc Cavazza, SC/TU (UK) Mario Vieira de Carvalho, FCSH-UN (PT) Né Barros, Balleteatro, Grupo Aesthetics, Politics and Arts /IF/ULP (PT) Nelson Zagalo, ICS-UM (PT) Nuno Correia, DI/FCT/UNL (PT) Nuno Jardim Nunes, UMA (PT) Patrícia Gouveia, ULHT (PT) Paul Brown, Sussex University (UK) Paulo Bernardino, DCA-UA (PT) Paulo Dias, IEP-UM (PT) Paulo Ferreira Lopes, EA-UCP (PT) Paulo Mendonça, EA-UM (PT) Pavel Sedlák, CIANT (CZ) Pedro Branco, DSI-UM (PT) Pedro Faria Lopes, ISCTE (PT) Penousal Machado, UC (PT) Perla Innocenti, HATII-U.Glasgow, (UK) Roberto Bresin, KTH (SE) Rosangella Leote SCIARTS(BR) (BR) Sandra Pauleto, U. York (UK) Sol Alonso Romera, UV (ES) Sudhir Mudur, UC (CA)! iii
7 ! ARTECH th International Conference on Digital Arts, 22 & 23 April, 2010 UM, Guimarães, Portugal Teresa Chambel, DI-FCUL (PT) Tomás Henriques, FCSH-UNL (PT) Valentina Nisi, Trinity College Dublin (IR) Steering Committee Adérito Marcos, UAberta (PT) (Chairman) Álvaro Barbosa, EA-UCP (PT) Christa Sommerer, K.U. Linz (AT) Henrique Silva, Bienal de Cerveira (PT) Leonel Valbom, ESGalllaecia (PT) Lola Dopico, FBA-UV (ES) Nuno Correia, DIFCT-UNL (PT) Seamus Ross, U.Glasgow (UK) Local Organization: Pedro Branco (Chairman) Bruno Figueiredo Miguel Duarte João Martinho Moura! Sponsoring and Supporting institutions iv!
8 ! ARTECH th International Conference on Digital Arts, 22 & 23 April, 2010 UM, Guimarães, Portugal Table of Contents Full Papers Creative Visualization and Exploration of Video Spaces Teresa Chambel, Telmo Rocha, and João Martinho 1 Methods in Digital Architectural Design - Understanding the Aspects that Enable a Re-active, Two-way Communication in the Digital Architectural Design Education Andrei Gheorghe 11 The Embodiment of Music/Sound Within an Intermedia Performance Space Caroline Wilkins and Oded Ben-Tal 19 Discerning the Cyborg Utterance; Alpha a Work for Electronic Sound Joanne Thomas 25 RTiVISS Real-Time Video Interactive Systems for Sustainability Mónica Mendes 29 Materializing Virtual Sites Linda Matthews and Gavin Perin 39 Dynamic Mapping Strategies for Interactive Art Installations: an Embodied Combined HCI HRI HHI Approach Pieter Coussement, Marc Leman, and Michiel Demey 45 Engaging Interactivity in Public Space Jiun-Jhy Her, Jim Hamlyn 52 Locative Narratives as Experience: A New Perspective on Location Aware Multimedia Stories Valentina Nisi, Ian Oakley, Martine Posthuma de Boer 59 Visualizing the Portuguese Empire Expansion and Decline Pedro Cruz and Penousal Machado 65 Hacking Media Education by Developing Interactive DIY- Whiteboards (mis)using Wiimote Daniela Reimann, Werner Fütterer, Sebastian Biefang 72 Revendo a Poética da Mobilidade e Espaços Híbridos Pervasivos Rosangella Leote 78 Das Geometrias aos Sistemas como Obra de Arte Andréia Machado Oliveira e Hermes Renato Hildebrand 85 THE APPROACH - Arte e Neurociência a Memória como Reconstrução Maria Manuela Lopes 94 Materialização e Virtualização nas Media Milton Terumitsu Sogabe 101 [NET]AR[G]Ts: Experiências Transmedia e Narrativas Cruzadas Patrícia Gouveia 108 Estudo para a Instalação de Unidade de Prototipagem Rápida Pedro Santos, Luís Caló, Lia Moreira, António Carvalho, and Gonçalo Furtado 118! v
9 ! ARTECH th International Conference on Digital Arts, 22 & 23 April, 2010 UM, Guimarães, Portugal O Papel Intuitivo do Desenho na Aprendizagem das Relações entre o Som e a Imagem José Miguel Gago da Silva 124 Espelhos Matriciais & Alter-eGo, duas Perspectivas Diferentes, Complementares ou Contraditórias da Identidade Renato Roque, Franscisco Calheiros, Rocio Garcia-Robles 132 La Coreografía Digital Interactiva Ludmila C. Martinez Pimentel 140 Visualización de Información: Diseño Gráfico y Lenguaje Digital Aberto J. García Ariza y Lola Dopico Aneiros 150 Arte e Design Digitais: O Uso da Retórica Procedimental como Fundamento da Discursividade Monica Tavares 150 Short Papers Poetics of Fiction: The Physiognomy of the Abyss João Ó 162 Sonata for Unprepared Player Tiago Videira 165 Visible and Audible Spectrums - A Proposal of Correspondence André Rangel Macedo 168 Epigenetics as Aesthetic Instrument in a Generative Virtual Ecosystem Rui Filipe Antunes, Frederic Fol Leymarie 172 Mirror Anamorphosis of 3D Object An Apparatus for Visual Data Transmission Marijana Kalabié 177 From Things To Burn: Inversed Metaphor To Painting After Technology: Digital Interactive Audiovisual Installation Producing a Analogical Data Visualization Painting Rudolfo Quintas 181 ArtDoc3D Digital Representation of Installation Artworks Ricardo Noguês and Nuno Correia 183 Conservation and Documentation of New Media Art. Italian theory and International strategies Laura Barreca 187 Playing with Cells, Artistic Installation Teresa Almeida, Inês Albuquerque 191 Saluting the Sun: Application of 3D Modelling to Representing the Human Figure R. Rawatlal 194 The Role of Different CAD Applications in Product Design Teaching Mário Barros 197 Laboratório Invisível : Onde Arte e Ciência se Encontram Rosana Horio Monteiro 201 Arte, Ciência e Tecnologia: Uma Relação Desafiante Para a Arte Contemporânea Inês Albuquerque, Rosa Maria Oliveira 204 vi!
10 ! ARTECH th International Conference on Digital Arts, 22 & 23 April, 2010 UM, Guimarães, Portugal A Sequenza III de Berio: Uma Experiência Multisensorial Isabel Nogueira e João Vilnei 207 Projecto Inside Sérgio Eliseu 210 Obsessão Imersiva Sérgio Eliseu 214 Criando no Escuro Tátil das Moléculas Anna Barros 218 Math4Kids Aprender Conceitos de Matemática Brincando David Jardim, Pedro Faria Lopes e Isabel Machado Alexandre 221 Art Installations Meridiend Michael Filimowicz 225 Stepping on the Light Michael Filimowicz 226 Over the limbo Chiara Passa 227 Cursor Caressor Eraser Michael Filimowicz, Melanie Cassidy, Andres Wanner 228 The walls have ears, 2009 Filipe Pais 229 I am not dead but I am divided - Theaterproject with constant interactive Web documentation Paul Wiersbinski 230 The Waterwalker (2009) Joaquin Gasgonia Palencia 231 Your life, our movie Fernando Velázquez, Bruno Favaretto & Francisco Lapetina, 232 Seeenergy Cristina Sylla, Ana Brandão Carla Martins, Manuel João Ferreira, Pedro Branco 233 PUSH COLOR - O ritmo da forma Vera Bighetti 234 Galeria Pública para Artes Digitais / Public Gallery for Digital Art/Fénix" /2009 Silvestre Pestana 236 Starcity Alberto Ariza 237! vii
12 Creative Visualization and Exploration of Video Spaces Teresa Chambel 1, Telmo Rocha 1,2, and João Martinho 1 1 LaSIGE, Faculty of Sciences, University of Lisbon, Portugal 2 Portuguese Catholic University-School of Arts, Porto, Portugal Abstract This paper presents creative mechanisms to visualize and explore video spaces in 2D and 3D, with a semantic focus on cultural aspects, and stressing features such as color dominance, rhythm and movement, at the level of the video space and the individual videos, in increasingly rich and immersive experiences. These mechanisms were designed to allow to capture, search, experience, and express videos properties and relations, providing the means to gain new insights into our culture and to influence the expression of its intrinsic aesthetics in creative ways, at the crossroads of information access, culture and digital art. Index Terms Art, color, motion analysis, interactive systems, three-dimensional displays, video, visualization. I. Introduction Video is becoming a dominant medium in richer and broader widely accessed media spaces. By combining pictures, text and audio that change in time, video is in itself a very rich medium, thus providing huge amounts of information and an excellent platform for creativity to be expressed and explored. For instance, the pioneers of Video Art have been exploring this creativity, using installation and performance art in conjunction with video in order to create immersive experiences. However, the richness that makes video based information spaces so interesting, inside each video and outside in the information spaces where in so many ways they relate to each other, comes with a challenging complexity to handle, since video information is not structured. Visualization techniques could help to handle the complexity and express the richness in these information spaces, through intuitive and effective ways to convey meaningful information in the video . However, due to a lack of effective techniques to convey complex information intuitively through automatic video processing, an alternative approach is to provide an overview by extracting interesting information and presenting it in a meaningful way [5,10], for example through summarization based on properties like movement, rhythm or scene change. Also in Video Art, some of the works make use of visualization - as a tool to convey some kind of meaning to the viewer. We present mechanisms designed to support the interactive and creative visualization and navigation of videos, allowing to explore cultural relations among them and to experience and influence their aesthetic properties. These mechanisms have been experimented in environments that have been developed to explore 3D and 2D visualization and navigation inside and between video and video spaces, based on their semantic and lower level properties, like color and motion, and addressing cultural aspects in videos from different countries: Portugal, Spain and Brazil, in themes that include music and dance, from different authors or artists. This paper focuses in the creative visualization and interactive navigation aspects, providing a systematic perspective based on our previous work [20,27], new developments and insights. Section II presents mechanisms to visualize and explore the video space level, Section III addresses the representation of the videos in the video space level, while section IV focuses on visualizing and experiencing the actual videos. Section V presents search facilities that help to filter the videos in the video space into more manageable sets of videos that can be browsed and explored. Section VI takes a step further, addressing the mechanisms that support interactive creativity and painting with videos; and section VII addresses modalities explored to provide interactive and increasingly immersive experiences. Finally, section VIII presents and discusses a selection of related work; and the paper ends with conclusions and discussion of perspectives for future work in section IX. II. Visualizing and Exploring the Video Space At the video space level, several videos are presented from different perspectives. The challenge is to provide the means to visualize, organize and relate the different videos in significant ways to the user. At this level, the user can search, view, compare and interact with a selection of videos, through different views that emphasize different properties. Semantic properties capture meaning. In our examples they focus on the themes, like music and dance, their style, authors, and countries, and where captured through keyword annotation. Lower level properties include features like color dominance, rhythm and movement. In our examples these were captured through video processing and analysis. From the video space level, the user can also select and navigate to each one of the videos, to watch and experience them and to further explore their individual features. The following sub-sections describe different approaches to visualize and organize the video space. 1
13 A. 3D Globe View In this space view, videos are visualized on a 3D globe, adopting a world metaphor, organized in accordance to a chosen classification, for example their theme or country. Videos from the same category are linked together by a vibrant light bolt starting at the center of the globe - representing the user, that can be captured through a webcam - to the first video, then around its frame and to the next video. These links have a different color for each category, making it easier to visualize relations among videos. Videos are organized around the globe with vibrant light links, with an aesthetics that inspires movement and invites for action. The user can navigate this space by: spinning the globe in different directions and at different speeds; zooming in or out; moving inside and out, and changing perspectives by choosing a different category (theme, author, country, etc). Labels for the video properties in the chosen categories can be made visible or invisible by the user. From each video, the user may obtain additional information, or select it to set the focus and navigate to the individual video level. This video space view is exemplified mainly in Fig. 1 (left) and Fig. 2 (top). B. Grid View In this space view, videos are organized in a grid. In our example, videos from the same category are linked together by vibrant light bolts starting at the top - representing the user - to the first videos of each column, then around their frames and down to the next videos. This view gets videos side-by-side in accordance to the chosen classification making easier the comparisons among them. Fig. 1 (center) exemplifies the grid view, where music videos are organized by author. This view corresponds to a reorganization of the globe view, by getting the center of the globe to the top, and letting the linked videos of each category stretch down as if the gravity was pulling them down now. The user can change between these two views of the video space, performed as a smooth animation. Fig. 2 (bottom) exemplifies the transformation of a grid view to the corresponding globe view. C. World of Physical Particles In this view, the video space is presented in the form of a physical particle system where a collection of video icons, or views, move on the screen in accordance to their similarity in terms of a chosen property and the user interaction. Videos are represented as physical particles gliding on the screen, where the forces among the videos are dependent on the chosen property, attracting like and repelling different videos in accordance to that property. Therefore, these views also allow searching by similarity (see section V). The system has the particles starting at random positions and then evolving as the forces between them act. This type of view is exemplified with different video representations (presented in the next section): in Fig. 1 (right) with video loops; in Fig.3 (left) with videos most dominant colors, and (center) with videos dominant colors along time; Fig. 9 bottom (right) with video slit scans; and Fig. 10 (bottom) with videos dominant colors, as circles that start at random positions and attract similar and repel different colors. At the video space level, the user can cycle through different views, to get an overview of the videos contents in different perspectives, and select any video she chooses to visualize and experience at the video individual level, as presented is section IV. Figs 5 and 6 exemplify the access to the 3D video view from the 3D video space level. In the physical particles video space views, videos are accessed, also by user selection, through the integrated perspective view (Fig.3). III. Video Representations in the Video Space Video includes an enormous amount of information that changes along time. To represent videos in the video space we face the challenge of finding views that can show potentially long videos in a glimpse, at once or in a short time. For this, we need to explore ways to represent and summarize the videos. We present some of the views we explored for this purpose, stressing different perspectives or properties. A. Selected Still Image Frames In this approach, videos are represented by one still image that stands for the whole video. It is usually a frame selected from the video. This is a simple and widely adopted approach - YouTube and Bestiario interface for TED s videos , for example, use this type of video representation - but it falls short in capturing the video richness and even in suggesting that it is a video, and not only an image. B. Sampling Video in Loops Video loops include a sequence of video frames presented in indefinite cycles, for a content overview or summary. This is an effective way to show a summary and to suggest it is video, and not just a still image. Frames in traditional video loops are taken at constant time intervals, or sometimes at the beginning of each scene. The latter requiring more video processsing. In the 3D globe and grid views, videos are represented by loops (Figs 1 and 2). Figure 1 (right) presents a world of particles view, also with video loops. Average and dominant color loops are also video loops, but present selected frames based on their proximity to these color properties (explained in subsection C), providing a content overview that is more representative of the video colors. We created these 2
14 types of loops, but in a figure they would look like the ones presented for the more traditional loops, although the selected frames in the loops are not the same. C. Dominant and Average Colors For a stronger emphasis on the colors, videos can be represented by their average or dominant colors. The average color of a video is found by analyzing the colors present in the frames - mainly through color histograms - along the video, and calculating an average value; while the dominant colors are the ones that appear more often in the video frames . In these views, we chose to represent videos as circles filled with their average or dominant color (Fig. 10 bottom); or by striped rectangles, where the stripes feature video dominant colors with their width reflecting the proportion in which they dominate in the video (Fig.3 left). D. Dominant and Average Colors along Time As a dynamic media type, video changes along time, and so do the dominant and average colors. In these views, videos are represented by colored stripes showing the evolution of the average or dominant colors on each frame on a timeline like visualization. Fig.3 (center) exemplifies this view. In the video space, videos are represented by condensed versions of these views, but these can be expanded and scrolled through interactive onover selection. This is what is happening with the long purple stripe in the mid-right, which corresponds to the purple Sevillanas dance video that is presented to its right, in the same Fig. For a stronger emphasis on evolution along time, we considered motion aspects, presented in the next subsections. After all, video is made of colors in motion. E. Capturing Motion through Scene Averaging Motion of each scene can be captured in one single still image, having each pixel represent the average pixel color occurring in that same position in each frame in the scene (Fig. 7). By identifying scene changes and applying the same process to every scene in a video, we can obtain a sequence of frames that capture the motion along the video and present them in a loop. Scene average loops look like the view in Fig. 1 (right) but include scene averages (similar to Fig.7) instead of selected video frames. F. Capturing Motion with Slitscans Slit scans  capture, in a still image and in sequence, the action happening in the central area of the video. Instead of blurring the movement into frame sized images for the scenes that are presented in sequence, like in the average process, this method focuses in the central activity and expands across the duration of the video. In this view, scene changes can also be noticed by the user - through the discontinuities or abrupt changes in the horizontal sequence. In the video space (Fig. 9 bottom left), slit scans are presented in a condensed or summarized form that can be expanded and scrolled through user interaction over the image, as exemplified in the mid left for the purple Sevillanas dance video that is presented in Fig.3 (right). A close-up of this expanded slit scan is presented in Fig. 9 bottom (left). The top of the same Fig. presents slit scans from: the dance video (on the left) that was exemplified in the scene average view in Fig. 7; and a video that presents people talking, starting with a talking head close-up (on the right). Through slit scans, videos from different cultures will leave their own traces behind. All the presented video views are created using the info extracted in a previous video analysis phase . IV. Visualizing and Experiencing the Videos After accessing each video in the video space, the user wants to watch it, to know more about it and to experience it fully. The challenge here is to devise the means to provide the user with the information and an experience that is effective, enjoyable and potentially immersive. From here the user should also be able to access the video space to find about other videos that relate to this one in different perspectives. A. Video Playing The user can watch the full video and navigate it along the timeline with the traditional controls (not always visible, for aesthetic purposes). The video can be chosen from the video space, or captured from the real world by a webcam controlled by the user. The video playing is presented in the center of the integrated view in Fig. 3 (right), and at the top left corner of the 3D video view, for example in Fig. 4. B. Experiencing the Video in 3D In this view, a single video is represented in 3D, by pixels that take the shape of quadrangular prisms, in the original color, with their height reflecting brightness. As the video plays (Figs. 4-6, 10 top), pixels change color and height accordingly, brighter pixels standing taller and providing a 3D representation of the image. This was called the microspace level in this 3D space metaphor . The user can: increase or decrease resolution, by changing the number of pixels; enlarge or reduce the image, by changing the size of the pixels; spin the video in every direction, zoom in and out, navigate among and around the pixels, at adjustable speeds, having perspectives from the inside or the outside. In fast movements in or out, the video appears to explode or implode. It is possible to have the original video shown in small as a reference (in the top left corner in the Figs.), at the same time, or just stay with the video 3
15 representation. As the user gets more detached from the original video image, by changing pixel size or navigating inside the video, the original image gets deconstructed and the user is led to a more pure aesthetic experience of shapes, colors, light and motion. Video becomes an accidental source of aesthetic properties, influenced by the user. Figs. 4-6, and 10 top, exemplify this view through the visualization and navigation of music and dance videos. C. Visualizing Video through Integrated Perspectives Besides watching the video, this view allows accessing details and different views of the video. Its main purpose is to provide an integration of complementary perspectives that allows the user to get a more complete view, and also to serve as a portal to the video space through these different perspectives. Fig. 3 (right) presents the individual view of the purple Sevillanas s dance video. The video is playing in the center. In the top row, the color aspects are highlighted through the average color loop and circle (purple), followed by the dominant colors rectangle, and the dominant color circle (a brownish dark red) and loop. Note that the dominant color prevails in the striped rectangle, as expected. To the left of the video, the traditional loop, to the right, the average scene loop - stressing movement aspects - complemented by the scrollable slit scan in the bottom. D. Closing the Cycle: Accessing the Video Space from the Individual Videos The challenge here is to provide the means to access the video space from any individual video, allowing to reach videos that relate to this one in specific ways and through different perspectives. From the 3D view, in sub-section B, the user can navigate back to the video space, at any time, or select one of the pixels and drag it to the video space (called MacroSpace in this metaphor) icon, colored with the last searched color. This will trigger a search by color, resulting in a new presentation of the MacroSpace (Fig. 6). Through the integrated perspectives view, in subsection C (Fig. 3 right), users can access any of those views at the video space (through direct selection), where they can compare this video to the others, in the chosen perspective. For example, they can perceive different colors and rhythms in the dance videos from different authors and countries, in the slit scan view. V. Searching Videos Besides from the ability to navigate the video space through different perspectives that allow the user to find videos, some automated search facilities are provided. To complement the traditional keyword-based search, we explored search of videos based on their dominant or average colors. This can be done by: 1) Starting a search operation from any one of the video space views, and selecting one or more colors from a palette, or from the real world, through a webcam (Fig. 8 left). In the case of more than one color, percentages of dominance for each color can be specified. In any case, a threshold can be defined for the search precision. Selected videos are presented in the same view where the search was issued. 2) Selecting a pixel in the 3D video view, and dragging it to the MacroSpace icon (Fig.6). The resulting video space will present the selected videos organized by the same categories as the last time it was presented, but ordered from the center by average color similarity to the searched color. The center of the world also changes its color to the searched one every time. This way, the visual aesthetics of the MacroSpace is influenced by the user, and as a consequence, a new palette of videos is created to choose from and influence the next microspace aesthetics. This will be especially noted, if the videos have a clearly identified dominant color. The views based on physical particles also allow local searches by color similarity, either by just watching particles finding their way, either by interacting to reposition or change speed and direction of selected particles, changing the focus and influencing their path. VI. Interactive Creativity: Painting with Videos Creative processes often rely on chance as an important element, complementing user control, in the authoring of digital artwork [14,25]. We explored different processes that somehow deal with uncertainty and allow the user to influence the results, in ways that she can find, express and experience creative outcomes of meaningful and aesthetic qualities. A. At the Video Level Views like average scenes and slit scans can be seen as creative views of the video content, stressing properties like color and motion. The users can take the chance on individual video content, and control parameters like threshold to influence the aesthetics in the creation of these views . Fig. 7 presents average scenes capturing the color and motion of individual scenes on a music video, on the left, and a dance video, on the right, with an impressionist aesthetics. Fig. 9 presents slit scans. These can also be seen as paintings or photographs that capture motion. In the 3D view of the video, the user can experience and navigate the video in different ways, as explained in section IV B. This dynamic navigation already provides a creative outcome that is experienced by the users and possibly their audience, for example in Vjing scenarios, navigating to the sound of a music while watching the video in this manner. In addition, the user can also interactively, and in an invisible way, increase and 4
16 decrease the level of trace left behind when moving around, and the color hue of the background. This way, the interactive navigation can be captured in a painting like style, where the video keeps contributing with the dynamic and changing paint, either from a video chosen from the video space, or from the real world through the webcam (Fig. 10 top). The video brings some uncertainty, the users influence the outcome, through navigation, by changing trace level and background color, and possibly by influencing the pixels colors and heights through the movements they perform, or the colored objects they shows in front of the camera - creating a 3D dynamic painting in front of them. This was probably the feature most appreciated by the users that interacted with the VideoSpace installation . B. In the Video Space In the views based on physical particles, users can take the chance on the video properties, color search and the forces in the video space, and interfere by selecting the view that is presented, the videos that are on the set, and then influencing the movements and zooming to create the visual outcome. In these physical worlds, all the particles have the same mass and random speed at the start of the evolution. Once the particles are deployed, forces of attraction or repulsion are created among them, depending on their similarity. For example, similar colors tend to group and follow each other, while colors that are dissimilar will react and run away from each other, colliding on the screen boundaries and coming back. Since the world does not have any gravity, particles will bounce on any direction. This interaction among the particles eventually fades off after a while, because of a small drag force introduced and the tendency to get the repelling forces apart. This way, the movement does not become perpetual. Users can interact with the system by dragging the particles around, changing their speed and direction and then seeing the interactions produced among them. They can also zoom in and out on the particle system, changing the size and position of the particles, and choose to capture the trace the particles leave behind. This way, in a mixture of chance and user action, dynamic visual effects can be created and captured. Fig. 10 (bottom) exemplifies one of these creations based on the average color view of the video space, resembling an abstract painting. Note how similar colors tend to approach, and remember that the user can interfere with this tendency at any time. The results represent color dominance of videos in certain contexts (according to the selection made for the current video set in the video space) adding to a meaningful perspective of the information accessed. In the 3D globe view, if the users interactively change the level of trace left behind and background color, they can also capture the navigation around the world in creative aesthetic ways. Fig.2 captures this type of navigation in the globe view (top) and from the grid to the globe view. In all the reported situations, the interactive creative processes can be lived as ephemeral experiences, but their outcomes can also be captured at any time in the process, through screen capturing, available in the exemplified applications. VII. Interactive and Immersive Experiences Different interaction modalities are being explored to provide the user with effective information access and, in some situations, immersive experiences. Capturing the real world can also contribute to more immersive experiences and a close connection between the real and virtual worlds. A. Interaction Modalities All the interactions can be done with mouse and keyboard, and most of them can be done just through point, click and drag, also adequate for touch screens. Interaction through keyboard and mouse is adequate for most situations. Keyboard in particular can be the preferred mode for experienced users, for example in a VJing scenario, allowing better performance and accuracy by selecting the right keys right away. However, other interaction modalities, based on colors and motion for example, can provide more natural and deviceless interactions in an installation setting. In a first prototype of VideoSpace, we tried gesturebased interaction with an Hemisson [15,28] robot with eight sensors activated by proximity. But the accuracy and required proximity to the robot did not allow for a very natural interaction. In later versions, we experimented: with the detection of predefined colors in specific regions in front of the screen, having the user wear colored gloves or hold colored objects; and with the detection of gestures in large touch screens (Fig. 8 right). Light conditions are a challenge in the detection of colors; and a common challenge in both approaches is to find good and natural mappings between the colors and screen regions, or the gestures, and the operations to perform in the application. However, preliminary results are encouraging. The alternate interactions provide the flexibility to accommodate different types of users and scenarios of usage, towards more natural and potentially immersive experiences. B. Bridging the Gap Between Real and Virtual Worlds Besides the capturing of colors from the real world, for the purposes of searching by color at the video space level, and painting at the 3D video level, the user can in a certain sense get inside the video space. 5
17 In the video space globe view, the user is represented in the center of the world by a human figure, as the one responsible for defining the criteria to search and organize the video space, although she can actually navigate it more freely, as an observer, from any position in the space, like in any other view. In addition, this representation of the user is also a portal for their entrance into the video space. From here they access an individual video view, showing them in real-time and whatever they choose to present in front of the webcam. Besides from all the navigation described, the users may now have a direct influence on the video that is presented, and are actually being created by them from the real world in real-time, and influence the outcomes in the virtual world. VIII. Related work Most visualization tools and applications, found in surveys like  are used to visualize social networks and the internet, a few visualize music, the Amazon, and Flickr. Some of them are simply intended for artwork, others for performing exploratory data analysis. Video is not usually addressed. Exceptions include the visualization of videos in YouTube and Video Sphere. From each video on YouTube, the user can access a 2D view that represents videos as circular scattered still images, giving access to the traditional page to watch the video. It allows for visual neighborhood navigation, but provides limited functionality and information about the videos or the video space. VideoSphere  represents a video space around a 3D sphere, with links among the videos, reflecting semantic compatibility, and allowing navigating around and inside and out the sphere. When a video is selected, the sphere rotates to put it in the center, where it can be played. Links are represented in black by static lines, except for the links involving the video in the center, which become red. The user may also choose to hide the links, or to access a list of topics, and from there reach the videos related to each topic. Although more related to our 3D visualizations, it is restricted to the video sphere, without special support for the visualization of the videos other then still keyframes and traditional video play, with the focus on exploring semantic relations, in a less dynamic way. In terms of the support for search, the recent Multicolr Search Lab  finds still images based on selected colors. So far, it is available for Flickr and Alamy Stock Photography. This type of photo sharing sites usually allows searching based mainly on tags, titles and keywords, not colors. The same is true for video sharing sites like YouTube. We developed video search mechanisms that go in this direction. There are a number of works that explore new ways to view an individual video. In , Fels and Mase show a tridimensional view of the video by considering the video data to be a volume, where the third dimension is the time, which is similar to what Daniel and Chen present in . This kind of three-dimensional visualization provides the user with a way to view the evolution of a given area of the video throughout its duration by using the traditional x and y-axis as the width and height of the frame and the z-axis as the time, giving depth to the volume. Slit scan imaging techniques capture time-based phenomena into static images, being an interesting candidate to represent videos.  collects information about slit scan video artwork. Timeline  adopts slit scanning to allow easy and rapid exploration of a video history to view participants in collaborative scenarios. In  the authors make use of a comic book like presentation in order to summarize an individual video. The frames are clustered using the hierarchical agglomerative clustering technique in order to produce segments that can be represented by a keyframe, which is then presented. Although an interesting summarization, it is too large to represent videos in the video space. In a related approach, Irani et.al.  make use of the mosaic, or tile, based representation in order to present significant frames in a video sequence. In previous work, we explored this among other types of representation to index video and help the navigation in hypervideo . As mentioned before, their required dimension do not make them such good candidate for the video space, but they could be used to select the frames to include for example in a video loop to provide a good summary through video skimming . The Hitchcock editing system, forms video stacks organized by color histogram similarity, providing a visually perceptible way of organizing large collections of videos. Artistic representation and visualization of video has been addressed by a few authors. For example, Hertzmann and Perlin  developed methods for painterly video processing, applying paint only on regions where the source video is changing, producing video with a novel visual style. In a related approach, Litwinowicz  explores pushing short brush strokes along scene movements in video, providing tools for edition and correction. Artists express their creativity in ways intended to engage the audience s aesthetic sensibilities or to stimulate mind and spirit, sometimes into unconventional ways of seeing and perceiving the world. Digital art is created in a wide variety of approaches, contexts and experiences including: interactivity, non-linearity, multimedia, virtual and augmented reality, net communications, databases, and 3D visualization, often presented as art installations, where viewers may become participants in the art making process [9, 24]. The video art movement started around mid 1960 s and has been evolving with technology [8,21]. Recent video art works include entirely digitally rendered environments and video that responds to viewers movements or other 6
18 properties of the environment. Our visualization and navigation mechanisms are aligned with these properties and can provide support at different levels to the expression and experience of digital video art scenarios. I/O Brush  is a drawing tool developed at MIT Media Lab, to explore colors, textures, and movements found in everyday materials by picking up and drawing with them. It looks like a regular physical paintbrush but has a small camera with lights and touch sensors embedded inside. On the canvas, artists can draw with the special ink they just picked up from their immediate environment. Our views that allow webcam capturing, and in particular the 3D video view, where the user can influence the painting through the webcam, relate to this tool. However, instead of picking up the ink to draw with it, which closely matches the painting paradigm, in our approach, the user gets the ink as a continuous flow, allowing for more dynamic and immediate outcomes, in a new and hence less familiar paradigm. By providing natural forms of interaction, it also aligns with our goals in that direction. Creativity is also an important aspect in our work. There are a number of projects that use different and original approaches to the visualization and organization of videos and other types of information. For instance, in , Sims used genetic algorithms and the evolution theory in order to create graphics. In a related perspective, we also explored creative ways of editing videos, using evolutionary algorithms in our previous work . New video sequences were combined and selected, based on their characteristics represented as video annotations, either by defining criteria or by interactively performing selections in the evolving population of video clips. One of the objectives included exploring and discovering architectonic and cultural relations in videos from Brazil and Portugal, but we felt then the need for richer and more flexible ways to visualize and navigate video spaces. In the area of Video Jockeying (VJing) , performance artists create moving visual art, usually based on video on large displays, at events such as concerts, nightclubs, sometimes in conjunction with other performance art. Computational support for creative editing and visualization of video provide tools for these artists. As emphasized in previous sections, we provide interactive features that can be used by such an artist to present the videos in creative ways, in real-time, during the performance. IX. Conclusions and Perspectives We presented mechanisms for the interactive and creative visualization and exploration of videos in 3D and 2D video spaces, with emphasis on video semantic and lower level color and motion properties. Users can visualize the video space through different views, to search, compare and interact with a selection of videos in a cultural context, featuring areas such as music and dance from different authors and countries. Complementary views allow visualizing and emphasizing videos properties, helping to deal with their inherent complexity in more perceptive ways. Although built and experimented around a cultural theme, the mechanisms can be applied in many other contexts. This type of approach is becoming more important as video becomes increasingly pervasive. Video repositories like YouTube usually support tag based search, but visual properties like color and motion are usually not taken into account. After a previous filtering through automated search mechanisms, visualization techniques allow to present information in meaningful ways to ease the perception, comparison and search through browsing. The developed mechanisms can support users in different scenarios: the user may, for example, perceive and compare the different colors and motion patterns in Malhão and Sevillanas, folk dances from Portugal and Spain; finding a video from a band with a dominance of red and orange, along with other videos with similar color dominance, and to present them in creative ways in a certain video art project; allowing video jockeys to perform a selection of videos taking semantic and visual properties into account, and to let them interactively make a live performances that explore and expand on the aesthetic properties of the videos and the video space. These interactive environments can be used in a traditional setting with a screen, keyboard and mouse, or a touch screen, adequate in many contexts. However, we are also developing more natural and deviceless interfaces, adequate for installation settings and ambient interaction contexts. Along with the capturing of videos and colors from the real world, for data capture and search purposes, color and gesture based interactions are being tuned to allow for more immersive and richer user experiences  in the real and virtual spaces. As future perspectives, we intend to refine video visualization and search based on more systematic evaluations, complementing the preliminary encouraging results, and to explore new and integrated video properties, representations and views for the individual videos and video spaces in different contexts. In particular, we intend to complement the focus on color and motion properties with refinements to searches based on color and the inclusion of searches based on motion. For that, we need to process the videos in ways that capture this property, and to explore ways of expressing motion in interactive searches, either by descriptions or through user motion capture. In this direction, we also want to evolve further towards more immersive visualization and interaction techniques through the exploration of effective modalities and natural mappings between the interactions and the intended functionalities. 7
19 Finally, by following this work and finding synergies with our work in hypervideo , creative evolutionary video editing  and the classification and access of videos based on emotions , we would like to contribute further with new and effective ways to visualize, explore, and build video spaces that are creative and that inspire and support their users to express themselves and their own creativity. We believe we have designed creative, engaging and enjoyable interactive visual experiences based on some unconventional representations and expressions of video, and identified new perspectives to take the experience into new levels. Acknowledgement This work was partially supported by LaSIGE through the FCT Pluriannual Funding Programme. References  Bestiario, Videosphere, May  J. Boreczky, A. Girgensohn, G. Golovchinsky, and S. Uchihashi. An interactive comic book presentation for exploring video, in Proceedings of CHI '00, the ACM SIGCHI Conference on Human Factors in Computing Systems, The Hague, The Netherlands, pp ,  T. Chambel,,L. Correia, J. Manzolli, G.D. Miguel, N.A.C. Henriques, and N. Correia, Creating Video Art with Evolutionary Algorithms, Special Issue on "Technology and Digital Art", Computer & Graphics Journal, vol 31, issue 6, pp , Elsevier, Dec  T. Chambel, and N. Guimarães, Context perception in video-based hypermedia spaces, ACM Hypertext and Hypermedia, College Park, MD, USA, pp ,  G. Daniel, and M. Chen, Video Visualization, in Proceedings of the 14th IEEE Visualization 2003 (Vis'03). IEEE Visualization. IEEE Computer Society, Washington, DC, 54, October  S. Fels, and K. Mase, Interactive video cubism in Proceedings of NPIVM '99, Workshop on New Paradigms in Information Visualization and Manipulation, Kansas City, Missouri, USA, 78-82, November  S. Few, Data Visualization: Past, Present, and Future, IBM Cognos Innovation Center,  D. Hall, and S.J. Fifer (eds.), Illuminating Video: An Essential Guide To Video Art, Aperture,  E. Hatcher (ed.), Art as Culture: An Introduction to the Anthropology of Art, Bergin & Garvey,  A.G. Hauptmann, Lessons for the Future from a Decade of Informedia Video Analysis Research, International Conference on Image and Video Retrieval, National University of Singapore, Singapore, July 20-22, LNCS, vol 3568, pp.1-10, Aug  A. Hertzmann, and K. Perlin, Painterly rendering for video and interaction, Proceedings of the 1st International ACM Symposium on Non-photorealistic Animation and Rendering, Annecy, France, pp. 7-12, June 5-7,  Idée s Multicolr Search Lab,  M. Irani, P. Anandan, and S. Hsu, Mosaic based representations of video sequences and their applications, in Proceedings of the Fifth international Conference on Computer Vision. ICCV. IEEE Computer Society, Washington, DC, 605, June  G.F. Kneller. Art and Science of Creativity, International Thomson Publishing,  Kteam website, the Hemisson robot:  E. Lai-Chong Law, V. Roto, M. Hassenzahl, A. Vermeeren, J. Kort, Understanding, scoping and defining user experience: a survey approach, Proceedings of ACM CHI 2009, Boston, MA, USA, pp ,  G. Levin, and Collaborators Catalogues and Lists, An Informal Catalogue of Slit-Scan Video Artworks and Research,  P. Litwinowicz, Impressions of San Francisco. In Electronic Theater Program, nb.120 in SIGGRAPH Video Review,  M. Makela, LIVE CINEMA: Language and Elements. MA in New Media, Media Lab, Helsinki University of Art and Design, April  J. Martinho, and T. Chambel, "ColorsInMotion: Interactive Visualization and Exploration of Video Spaces", ACM Academic MindTrek'2009: Everyday Life in the Ubiquitous Era, Ambient and Ubiquitous Media track, Tampere, Finland, Sep-Oct,  Nam June Paik Official Website  M. Nunes, S. Greenberg, S. Carpendale, and C. Gutwin, What did I miss? Visualizing the past through video traces, in Proc ECSCW'07 European Conf on Computer Supported Cooperative Work, Springer-Verlag,  E. Oliveira, and T. Chambel, "Emotional Video Album: getting emotions into the picture", emotion-in-hci'2008, The 4th Workshop on Emotion in Human-Computer Interaction, at HCI'2008, the 22nd BCS HCI Group conference on HCI, Liverpool, UK, September,  C. Paul, Digital Art, London: Thames & Hudson,  R. Pepperell, Computer aided creativity: practical experience and theoretical concerns, In Proceedings of the 4th Conference on Creativity & Cognition, Loughborough, UK, 50-56, October  S. Perez, The Best Tools for Visualization, r_visualization.php  T. Rocha, and T. Chambel, "VideoSpace: a 3D Video Experience", Installation, in Proceedings of Artech'2008, 4th International Conference on Digital Arts, pp , Portuguese Catholic University, Porto, Portugal, November,  K. Ryokai, S. Marti, and H. Ishii, I/O brush: drawing with everyday ojects as ink, Proceedings of ACM CHI 2004, Vienna, Austria, pp ,  K. Sims, Artificial evolution for computer graphics, in Proceedings of the 18th Annual Conference on Computer Graphics and Interactive Techniques, SIGGRAPH '91, ACM, New York, NY, ,
20 Fig. 1. Video space views: (left) - globe view; (center) - grid view; (right) - world of physical particles, with video loops. Background color can be changed interactively by the user. Fig. 2. Navigating the video space in 3D: (top) - rotating and zooming; (bottom) - navigating from grid to globe view. Fig. 3. Video space and video views: (left) world of particles representing video space with videos most dominant colors; (center) - same video space with dominant colors along time; (right) - view with integrated perspectives of a single video. From here, the users can navigate to the video space in the perspective they choose. Fig. 4. Navigating a video in 3D: a music video by Luís Represas with dominant color blue. Users can zoom in and out, rotate, explode, implode, etc. influencing the video s aesthetics of colors and brightness in motion. 9